There are a variety of different loads and faults in all the many types of facilities that emergency power system designers encounter. Specifying an uninterruptible power supply system calls for a careful consideration of what is unique to the project.
CSE: How does one decide that UPS is even the best emergency power solution for a specific application?
ALBERT: In general, if any system or application is affected or operationally impaired because of power source interruptions or loss, UPS is required.
McCALLA: Any business processes that depend on technological systems need to be protected by UPS. Not only do UPS systems provide ride-though during outages, but also protect sensitive electronics from damage or degraded performance caused by sags, surges and other power quality problems.
REED: The best analysis starts with a look at what's called the "CBEMA" or "ITI" curve. The required input voltage envelope for the critical equipment, along with the power quality environment of the facility, determines whether there is a requirement for a UPS system.
I would think that most critical processing facilities nowadays have equipment such as servers, computers, VIP and automation systems and the like that would warrant backup by a UPS system.
CSE: How does one determine the best UPS system configuration for riding through blackouts and brownouts?
McCALLA: It starts with analyzing an owner's business requirements, and determining how much disruption and downtime would cost the business. Of course, the assessment changes as new applications are added, so it is important to think about future requirements.
TARDY: Because of the variety of different loads and types of faults in a facility, there are several considerations. But the major decision is usually whether to specify a centralized UPS or go with application-specific distributed UPS. To determine the type required, consider the following:
Are the loads non-essential or critical?
What equipment within the facility might generate noise, harmonic distortion or other power anomalies?
What are the power requirements: total power, input voltage, single- or three-phase?
What is the space availability: footprint, access, doorway sizes and raised floor?
Are there any environmental restrictions: temperature, humidity, ventilation?
How difficult is installation: special rigging requirements, access, etc.?
What are the maintenance requirements?
What is the life-cycle cost of the system?
ALBERT: Getting down to specific system considerations, line-interactive and online technologies provide voltage regulation, in addition to backup power. Consideration should also be given to required power capacity, backup time and transfer time, particularly if line-interactive is specified. Online, double-conversion UPS is recommended for sensitive equipment and critical applications.
REED: On the data center side, new concepts in reliability-centered maintenance infrastructure are affecting UPS configuration, specifically, establishing what are reliable criteria that directly relate the UPS system cost to the cost of an outage to the critical equipment. Does the UPS system need to be concurrently maintainable while the critical equipment is online? How many hours of availability per year does the facility require? Does the cost of an outage require that fault-tolerant UPS infrastructure be provided? Fault tolerant design provides the ability of the infrastructure to sustain at least one worst-case unplanned failure event without impacting the critical load.
Once the required reliability, availability and fault tolerance are defined—ANSI/TIA Standard 942 "Telecommunications Infrastructure Standard for Data Centers" can provide a base level of guidance—then the system configuration can be determined within the power system, building space and budget requirements.
CSE: What are some common mistakes in specifying UPS and how can the pitfalls be avoided?
ALBERT: A common mistake concerns load type. Each UPS technology is suited to a particular application, and loads that are highly inductive should not be powered by a UPS. In addition, the spec must consider total load. Always size the UPS for both the total volt-amperes and total watts based on future growth projections.
McCALLA: Failing to adequately plan for growth is definitely a mistake. For example, companies are reevaluating their power protection strategies as they implement Voice-over-Internet Protocol (VoIP), because the cost of failure increases when voice and data are on the same network.
If the UPS initially installed is too small, additional UPS modules will be required every time computing capacity is added. Many small modules means greater chance of failure and are more difficult to manage.
It's also important to develop a plan for adapting to increased requirements, particularly if a single module system is being installed. Adding redundancy to the UPS system allows the system to support higher levels of availability.
REED: Engineers often start with a "base" redundancy level and adjust to owners' budgets or other fixed scenarios. But this hinders proper analysis. Overestimating the initial watts per sq. ft. of critical load will result in a first cost, capital-intensive project that will be severely underutilized at move-in. Conversely, underestimating the eventual watts per sq. ft. load density on the raised floor environment can leave the owner with a facility that quickly becomes over-cabled, under-cooled and incapable of meeting the power requirements of the next generation of computing equipment.
Modularizing UPS and electrical systems so that additional capacity can later be added is a good strategy.
TARDY: Absolutely. Oversizing the UPS based on power supply ratings instead of actual power consumption is a common mistake. The typical spread is one to two, meaning that most UPS only run at 50% of their rated capacity. This can adversely affect UPS efficiency, which typically decreases with the load level.
CSE: Any other important issues to watch for in designing UPS systems?
REED: Maintaining physical separation between redundant equipment and redundant distribution paths. This is important so that a catastrophic failure, due to fire or water, at one location does not cause a single point failure of redundant systems.
Also, pay attention to the most critical UPS components. This often means a battery system. Whether valve regulated dry cells or wet cell types, there is a requirement to specify high quality batteries and to verify a quality installation by commissioning and field load testing. An often-overlooked issue is obtaining baseline data for future predictive trending of battery condition.
ALBERT: Think about total backup time. It is important to determine if the UPS will need extended runtime capability.
Also, a maintenance bypass switch allows UPS to be taken out of service for maintenance or replacement while keeping the critical load up and running.
Finally, make sure that UPS can handle the frequency and voltage range of the generator.
CSE: What are the latest offerings in UPS technology?
McCALLA: Advanced features include control systems, more compact designs, increased efficiency and a greater ability to reconfigure systems on site as needs change.
TARDY: Monitoring capabilities. IT administrators can quickly see all UPS across their entire network and remotely monitor and manage the devices—even over the Internet. Today's power management software assures automatic orderly shutdown of unattended servers during extended outages, automatic reboot and remote monitoring, as well as alarm condition notification via email, fax and pager.
REED: New innovations I've seen include transistorized front-ends (rectifiers) that provide high power factor and low total harmonic distortion numbers without the use of expensive and problematic capacitors or filters. Also, integrated UPS and standby generator systems, with flywheel or induction coupling support, is another newer offering.
ALBERT: I would add to that list scalable, redundant and built-in intelligence to provide a fault-tolerant system. Modular design, as Tom noted, also lends scalability to users as their power demands grow. Additional power or battery modules can extend the back-up time or add system redundancy. Also, hot-swap capability makes it easy to add or replace modules. Essentially, systems can be upgraded without re-investing in a new system and microprocessor controls perform self diagnostics tests that increase UPS functionality and reliability.
CSE: Tom, you noted flywheel technology? Can you expound on that?
REED: The design lifespan of a flywheel system is typically 50% greater than a 20-year flooded cell system. Maintenance can be performed every six years on the flywheel bearings vs. every three to six months for battery systems, as batteries require HVAC systems to maintain a relatively constant temperature band. Eliminating batteries also eliminates spill containment, disposal issues and EPA regs.
McCALLA: Flywheel systems are particularly gaining acceptance in high availability applications where they are used to enhance battery life and in applications where fast-start generators are already installed. When you can talk to customers about ways to either eliminate batteries or increase the reliability of batteries—whether it's flywheel systems or battery monitoring—they're very interested.
TARDY: Flywheels are also coming on to the scene for niche applications that don't require extended battery backup time. For applications where gensets are mandatory, UPS systems that use flywheels can be a good solution, as they can bridge the power transfer to generator. However, they are limited to about 15 seconds of hold-up time. For especially critical applications, flywheels can be used with batteries to maximize uptime and, by reducing battery charge discharge cycles, they extend battery life.
Thomas E. Reed , P.E. , Senior Director, Managing Partner, Critical Technology, Kling, Philadelphia
Suzette M. Albert , Product Marketing Manager, EGS, Sola/Hevi-Duty, Rosemont, Ill.
Kevin McCalla , Director of Marketing Uninterruptible Power Systems, Liebert Corporation Columbus
Herve Tardy , Vice President of Marketing for MGE UPS, Systems North America, Costa Mesa, Calif.